How can I explore tSNE/UMAP plots?
One of the goals of exploring a tSNE/UMAP map (as with any other dimensionality-reduced datasets) is to characterize the populations defined by it. In cytometry, this means defining the phenotypes of populations/”islands” defined on the plots to give them biologically-relevant naming. FCS Express allows users to efficiently explore and name populations in multiple ways: classical backgating, parameter overlays and density plots.
One of the most important steps in high-dimensional data analysis is the exploration of the dimensionality-reduced dataset obtained after tSNE, UMAP or other dimensionality-reduction algorithms. One of the aims of this step is the biological classification of the different populations mapped on the map. In FCS Express this process is extremely easy and dynamic.
Some options and ideas for visualization and analysis are presented below.
After creating a gate around population of interest on the dimensionality-reduced dataset (e.g. tSNE map, UMAP map, PCA, map,...) the population of interest can be backgated on classical 2D Color Dot Plots.
The gate of interest can then be moved around on the map allowing you to immediately see the population of interest backgated on Color Dot plots which will update in real time.
Tip 1: the NxN Multiplots feature can be used to quickly create an NxN matrix of plots.
Tip 2: the population of interest can be emphasized to make its identification easier on Color Dot Plots which is especially useful when the population is rare.
The phenotype of a population of interest can also be investigated using Parameter Overlays. The parameter overlay feature allows users to plot the distribution of the parameters on an overlaid histogram (a separate overlay for each parameter) that may be gated on population of interest defined on the map. By utilizing parameter overlays users can quickly assess the distribution of many markers across many parameters within a gated population.
Tip: Overlay can be easily customized via the Overlay Formatting dialog.
Event on Density Plots are usually colored based on the density distribution of cells within the plot itself (counts). The density heat mapping can also be representative of the heat of any parameter in the data set. For example, events on a density plot can be colored based on the expression level of a marker/parameter of interest and by visualizing many density plots representing many parameters, researchers can gain insights on which "islands" contain phenotypes of interest and monitor changes between groups.
FCS Express allows users to quickly and easily create and define groups of density plots that will heat map parameters of interest manually on one or few density plots or automatically via the NxN Color by All option.
FCS Express on Mac
Upgrading FCS Express